A few days ago I updated from Fedora Workstation 38 to Fedora Workstation 39. However, there was a big problem: I could no longer log in. Every time the login screen came up and I entered and confirmed my password, it loaded, but then I was thrown back to the login screen.
When I selected X11, however, it worked.
So I tried to start wayland from the console with str+alt+f5 with the command:
dbus-daemon[5378]: [session uid=0 pid=5378] Activating service name='org.gtk.vfs.Daemon' requested by ':1.0' (uid=0 pid=5379 comm="gnome-shell --display-server --wayland" label="unconfined_u:unconfined_r:unconfined_t:s0-s0:c0.c1023")
dbus-daemon[5378]: [session uid=0 pid=5378] Successfully activated service 'org.gtk.vfs.Daemon'
libmutter-Message: 16:43:02.047: Running GNOME Shell (using mutter 45.1) as a Wayland display server
libmutter-Message: 16:43:02.103: Made thread 'KMS thread' realtime scheduled
libmutter-Message: 16:43:02.106: Device '/dev/dri/card1' prefers shadow buffer
libmutter-Message: 16:43:02.111: Added device '/dev/dri/card1' (amdgpu) using atomic mode setting.
libmutter-Message: 16:43:02.113: Created gbm renderer for '/dev/dri/card1'
libmutter-Message: 16:43:02.113: Boot VGA GPU /dev/dri/card1 selected as primary
** Message: 16:43:02.115: Obtained a high priority EGL context
** Message: 16:43:02.116: Obtained a high priority EGL context
dbus-daemon[5378]: [session uid=0 pid=5378] Activating service name='org.a11y.Bus' requested by ':1.3' (uid=0 pid=5379 comm="gnome-shell --display-server --wayland" label="unconfined_u:unconfined_r:unconfined_t:s0-s0:c0.c1023")
dbus-daemon[5378]: [session uid=0 pid=5378] Successfully activated service 'org.a11y.Bus'
libmutter-Message: 16:43:02.392: Using public X11 display :0, (using :1 for managed services)
(gnome-shell:5379): libmutter-WARNING **: 16:43:02.392: WL: error: XDG_RUNTIME_DIR is invalid or not set in the environment
(gnome-shell:5379): libmutter-ERROR **: 16:43:02.392: Failed to create socket
== Stack trace for context 0x55d91c16b0f0 ==
A connection to the bus can't be made
Unfortunately I have not much experience and it would be grateful for any help or corrections if I have done something wrong
If it helps, here are some information about my device:
I can confirm this. The same happend to me while updating from FC38 to FC39. I am on a Intel i5 12th generation, ASUS B660 mainboard with a NVIDIA 3060Ti card.
Initially I thought, this might be a Gnome extension problem, but deactivating extensions proved that this is not the case. It seems like Wayland sessions are crashing under certain conditions.
Looking for help as well. Please let me know, if I can assist in testing a possible solution.
Thank you in advance!
I think, as far as I could tell, this was more of a Kde problem. I tried something similar on gnome once, but I had no success.
In general, though, I’ve also noticed that the error is slightly different, as I’ve realized that it’s some Wayland issue with Fedora 39 - at least I think so.
The thing that would help me is if you can’t log in at the login screen either, press ctrl+alt+f5 to get to the console and run gnome + wayland manually once and see if it gives the same errors
(With this command):
dbus-daemon[13532]: [session uid=0 pid=13532] Activating service name='org.gtk.vfs.Daemon' requested by ':1.0' (uid=0 pid=13533 comm="gnome-shell --display-server --wayland" label="unconfined_u:unconfined_r:unconfined_t:s0-s0:c0.c1023")
dbus-daemon[13532]: [session uid=0 pid=13532] Successfully activated service 'org.gtk.vfs.Daemon'
libmutter-Message: 11:37:27.412: Running GNOME Shell (using mutter 45.1) as a Wayland display server
libmutter-Message: 11:37:27.459: Made thread 'KMS thread' realtime scheduled
libmutter-Message: 11:37:27.459: Device '/dev/dri/card0' prefers shadow buffer
A connection to the bus can't be made
This is a different bug and I think it’s a nvidia-bug.
The next entry for me is about amd-gpu. If I understand correctly you are using nvidia and it is known to have problems with wayland (or I atleast heard from buggs with wayland and nvidia).
If I have understood correctly, there are problems between the Wayland compositor and your graphics.
As I said, I don’t know much about it, but I would try to update, install or reinstall nvidia-driver or something like that
If it is related to nvidia, the latest nvidia driver from rpmfusion is 535.129.03. If you are not using that version then please do an upgrade (kernel 6.5.11 is also available) then test again.
Thank you very much for your advise.
The system is on the latest kernel 6.5.11 and also on latest NVidia drivers.
Wayland worked fine before on FC38.
I will double check and continue testing next week.
Have a good weekend!
Thomas
dbus-daemon[6062]: [session uid=0 pid=6062] Activating service name='org.gtk.vfs.Daemon' requested by ':1.0' (uid=0 pid=6063 comm="gnome-shell --display-server --wayland" label="unconfined_u:unconfined_r:unconfined_t:s0-s0:c0.c1023")
dbus-daemon[6062]: [session uid=0 pid=6062] Successfully activated service 'org.gtk.vfs.Daemon'
libmutter-Message: 23:12:37.927: Running GNOME Shell (using mutter 45.1) as a Wayland display server
libmutter-Message: 23:12:37.979: Made thread 'KMS thread' realtime scheduled
libmutter-Message: 23:12:37.980: Device '/dev/dri/card1' prefers shadow buffer
libmutter-Message: 23:12:37.980: Added device '/dev/dri/card1' (nvidia-drm) using atomic mode setting.
libmutter-Message: 23:12:38.009: Device '/dev/dri/card0' prefers shadow buffer
libmutter-Message: 23:12:38.105: Added device '/dev/dri/card0' (i915) using atomic mode setting.
libmutter-Message: 23:12:38.107: Created gbm renderer for '/dev/dri/card1'
libmutter-Message: 23:12:38.108: Created gbm renderer for '/dev/dri/card0'
libmutter-Message: 23:12:38.108: Boot VGA GPU /dev/dri/card0 selected as primary
** Message: 23:12:38.109: Obtained a high priority EGL context
** Message: 23:12:38.110: Obtained a high priority EGL context
dbus-daemon[6062]: [session uid=0 pid=6062] Activating service name='org.a11y.Bus' requested by ':1.3' (uid=0 pid=6063 comm="gnome-shell --display-server --wayland" label="unconfined_u:unconfined_r:unconfined_t:s0-s0:c0.c1023")
dbus-daemon[6062]: [session uid=0 pid=6062] Successfully activated service 'org.a11y.Bus'
libmutter-Message: 23:12:38.284: Using public X11 display :0, (using :1 for managed services)
(gnome-shell:6063): libmutter-WARNING **: 23:12:38.284: WL: error: XDG_RUNTIME_DIR is invalid or not set in the environment
(gnome-shell:6063): libmutter-ERROR **: 23:12:38.284: Failed to create socket
== Stack trace for context 0x55e55c364760 ==
A connection to the bus can't be made
I have a NVIDIA GPU, but the kernel and rpmfusion are up-to-date (535.129.03), here is the info about my system:
I can contribute small findings about my Wayland problems: As all software is up to date and I have no clue on how to avoid these Gnome/Wayland/NVidia issue, I have tested a few things after updating my FC39 completely today.
Logging in under Gnome/Wayland still does not work on my user account on FC39.
Exactly the same configuration worked for years until the latest updates of FC38.
Logging in using Gnome/X11 under FC39 works flawlessly.
Next I have created a new user account and what a surprise: The Gnome/Wayland session is working fine using this new and unused user account!
My assumption: Some of the .config/* data is causing the termination of the Wayland session.
Can anybody reproduce this?
Does anybody have a hint on where to find trouble spots in the user configuration?
I do not see other differences at the moment between the two login users…
Thanks for sharing this.
I can reproduce it. Using a new profile I can login under wayland, but not using the old one.
Unfortunately I don’t know how to fix it…
Hunting for the setting which might hinder the start of a Wayland session on FC39 was not successful until now, but I have found another strange behavior with partial success:
I opened up Gnome Tweaks (“GNOME Optimierung” in the German user interface) and have reset the settings to default by means of the corresponding menu entry.
Now I need to login twice: First attempt most often fails, but the second attempt is successful. The Wayland session starts:
This is not a sattisfying solution and far from robust, but the first time I can start Gnome/Wayland using my account. This does not feel very relyable and I needed to reset the Gnome Tweaks again. Thus this solution does not seem to be persistent.
Did anybody else find a superior solution?
Just a remark on my system configuration: I am using a dual screen setup.
Is this a pattern for this failure? Or to be more direct: Are youalso running a dual screen system while observing this problem?
Can you compare the kernel command line from before and after the upgrade? Maybe the upgrade messed with options needed for nvidia or maybe nvidia now needs different options.
I am having a similar issue with an Nvidia 1650 Super, and I am also using Gnome Tweaks, but do not have dual screen. I have been on Fedora 39 since the beta. Issue actually started after upgrading to latest kernel and nVidia drivers.
I can log in, but have random Wayland crashes bringing me back to GDM.
Watching this thread.
Addendum: My Gnome Tweaks, like “Applications” and “Places” along the top taskbar have disappeared on their own. (I did not reset to defaults)
I followed your suggestion and “Reset to default” in Gnome Tweaks, but it did not allow me to log in with my default account under Wayland, even after repeatedly trying to enter the password. I even de-installed Gnome Tweaks, but this didn’t work either.
Something else that I tried (and that did not work) is to de-install the current NVIDIA driver and install the NVIDIA 470xx driver instead.
Even before the upgrade to F39 I occasionally experienced some weird behavior, usually after system updates: the first login attempt would fail, but the second one would work, however GNOME extensions were deactivated. Back then I was not bothered too much and did not link it to Wayland.
Concerning the screen setup: usually I use the laptop screen only, but once in a while I connect it to an external monitor. It doesn’t seem to make a difference.
hi guys,
I think I have a similar problem.
(my english is too bad, this post was translated with DeepL)
my configuration :
OS : F39
Laptop : LENOVO LEGION
GPU 1 : AMD Radeon,
GPU 2 : Nvidia RTX 3070
with two exteral scrreens
First of all, i had no problem when u was with F38.
when I only use the laptop screen or a single external screen, everything is OK.
But when I use the 2 external screens :
When I connect, window display is very slow (about 2 seconds lag). On the other hand, the mouse cursor moves without lag.
If I close the session and reconnect (with the same account, everything is normal.
On 1st startup (with slow display), if I unplug the HDMI/ISB-C cable from one of the two external screens, the display returns to normal.
I think it’s a problem linked to the NVIDIA driver (in association with the last 2 Kernels) since, unless I’m mistaken, it’s the Nvidia card that manages the display of the external screens on my laptop. Maybe we just need to wait for updates.
More data (affected hardware and software configurations, error logs) makes it easier for developers to fix bugs. Developers have access to a limited range of configurations, so may not be able to reproduce the issue without very specific details. I’m busy at present, and only use nvidia when I have an issue with nouveau, so have not systematically analyzed reports of Nvidia issues, but I wonder if the problems occur with fresh installs or only with updates.
AFAIK, the kernel options are located in the grubenv binary file. Correct? There is only one file with one set of options for every kernel. Is this correct? The file had been updated, thus it could be a source of the problem. But how can I analyze this file?
Surprise, surprise! This is not a solution yet, but here are my latest findings:
As “news news” wrote, it seems to be a dual screen problem.
I had disconnected the second screen (4k TV through HDMI, 2160p) and now only use my monitor (1440 p through DP) and the Wayland session is starting quite relyable!
I will keep watching this.
Will this help finding a solution? Please let me know, if I shall test anything in addition.
!